I took a slam course (14 Lectures on Visual SLAM) from Xiang Gao. I read through the book and followed this author to recode from scratch what he has achieved in this repo "https://github.com/gaoxiang12/slambook", which includes feature detection and matching, PnP, ICP and optimization(bundle adjustment) etc. Finally integrate above to develop a visual odometry demo(camera pose estimation) for TUM dataset freiburg1_xyz.
IDE: KDEVELOP.